Integrative genomic testing of cancer survival using semiparametric linear transformation models
- 16 February 2016
- journal article
- research article
- Published by Wiley in Statistics in Medicine
- Vol. 35 (16) , 2831-2844
Abstract
The wide availability of multi‐dimensional genomic data has spurred increasing interests in integrating multi‐platform genomic data. Integrative analysis of cancer genome landscape can potentially lead to deeper understanding of the biological process of cancer. We integrate epigenetics (DNA methylation and microRNA expression) and gene expression data in tumor genome to delineate the association between different aspects of the biological processes and brain tumor survival. To model the association, we employ a flexible semiparametric linear transformation model that incorporates both the main effects of these genomic measures as well as the possible interactions among them. We develop variance component tests to examine different coordinated effects by testing various subsets of model coefficients for the genomic markers. A Monte Carlo perturbation procedure is constructed to approximate the null distribution of the proposed test statistics. We further propose omnibus testing procedures to synthesize information from fitting various parsimonious sub‐models to improve power. Simulation results suggest that our proposed testing procedures maintain proper size under the null and outperform standard score tests. We further illustrate the utility of our procedure in two genomic analyses for survival of glioblastoma multiforme patients. Copyright © 2016 John Wiley & Sons, Ltd.Keywords
Funding Information
- National Institutes of Health (R03CA182937, R01AG048825)
- National Institutes of Health (R03CA182937, R01AG048825)
This publication has 33 references indexed in Scilit:
- iBAG: integrative Bayesian analysis of high-dimensional multiplatform genomics dataBioinformatics, 2012
- DNA methylation differences at growth related genes correlate with birth weight: a molecular signature linked to developmental origins of adult disease?BMC Medical Genomics, 2012
- Kernel Machine Approach to Testing the Significance of Multiple Genetic Markers for Risk PredictionBiometrics, 2011
- SZGR: a comprehensive schizophrenia gene resourceMolecular Psychiatry, 2010
- Finding the missing heritability of complex diseasesNature, 2009
- Phase 1/2 study of the combination of 5-aza-2′-deoxycytidine with valproic acid in patients with leukemiaBlood, 2006
- Radiotherapy plus Concomitant and Adjuvant Temozolomide for GlioblastomaNew England Journal of Medicine, 2005
- Semiparametric regression analysis for clustered failure time dataBiometrika, 2000
- Automatic Smoothing of Regression Functions in Generalized Linear ModelsJournal of the American Statistical Association, 1986
- Cox's Regression Model for Counting Processes: A Large Sample StudyThe Annals of Statistics, 1982